Quality report of a bibliography

Bibliography Setup

The bibliography is loaded and the DOIs, ISBNs and ISSNs are extracted. In a second step, the corresponding works are downloaded from [OpenAlex(https://openalex.org)].

Show the code

#|

bib <- read.csv(params$bibliography)

dois <- bib$DOI
names(dois) <- bib$Key
dois <- dois[dois != ""]

dois <- dois |>
    gsub(pattern = "^https://doi.org/", replacement = "") |>
    gsub(pattern = "^https://dx.doi.org/", replacement = "") |>
    gsub(pattern = "^https://hdl.handle.net/", replacement = "") |>
    gsub(pattern = "^http://doi.org/", replacement = "") |>
    gsub(pattern = "^http://dx.doi.org/", replacement = "") |>
    gsub(pattern = "^http://hdl.handle.net/", replacement = "") |>
    gsub(pattern = "^doi:", replacement = "") |>
    gsub(pattern = "^DOI ", replacement = "")

isbns <- bib$ISBN
names(isbns) <- bib$Key
isbns <- isbns[isbns != ""]

issns <- bib$DOI
names(issns) <- bib$Key
issns <- issns[issns != ""]

In a second step we retrieve the works based on the DOIs fromOpenAlex.

Show the code

#|

fn <- file.path(".", "data", "bib_works.rds")
if (file.exists(fn)) {
    bib_works <- readRDS(fn)
} else {
    doi_chunks <- split(unique(dois), ceiling(seq_along(unique(dois)) / 199))
    #
    bib_works <- pbmcapply::pbmclapply(
        seq_along(doi_chunks),
        function(i) {
            oa_fetch(
                entity = "works",
                doi = doi_chunks[[i]],
                per_page = 200,
                verbose = FALSE
            )
        },
        mc.cores = params$mc.cores
    ) |>
        do.call(what = rbind)

    saveRDS(bib_works, fn)
}

dois_oa <- bib_works$doi |>
    gsub(pattern = "^https://doi.org/", replacement = "") |>
    gsub(pattern = "^https://dx.doi.org/", replacement = "") |>
    gsub(pattern = "^https://hdl.handle.net/", replacement = "") |>
    gsub(pattern = "^http://doi.org/", replacement = "") |>
    gsub(pattern = "^http://dx.doi.org/", replacement = "") |>
    gsub(pattern = "^http://hdl.handle.net/", replacement = "") |>
    gsub(pattern = "^doi:", replacement = "") |>
    gsub(pattern = "^DOI ", replacement = "")

Data Quality of the Bibliography

Cleanliness of bibliography

One measure of the cleanliness of a Bibliography is assessed by checking the number of references that have a DOI. The following table gives an overview over some numbers regarding the DOIs, ISBNs and ISSNs in the bibliography.

Entries with DOIs, ISBNs or ISSNs

To identify a reference, the most widely used identifier is the DOI. The following table shows the number of references with a DOI and the number of unique DOIs.

To consider duplicate ISBNs or ISSNs as duplicates entries in the library is not waranted as e.g. differenc chapters of a book can be separate entries in the library and therefore lead toi duplicates.

DOIs: 3863 (73.8482126%) - 2 duplicates
ISBNs: 453 (8.6599121%) - 57 duplicates
ISSNs: 3863 (73.8482126%) - 2 duplicates

The following DOIs are duplicates in the bibliography. This table should be empty.

Show the code

# duplicate_isbns <- paste0("https://isbnsearch.org/search?s=", dois[duplicated(isbns)])
# duplicate_issns <- paste0("", dois[duplicated(issns)])


data.frame(
    Type = "doi",
    Identifier = sprintf('<a href="https://doi.org/%s" target="_blank">%s</a>', dois[duplicated(dois)], dois[duplicated(dois)])
) |>
    knitr::kable(
        caption = "Duplicate DOIs in the Bibliography",
        escape = FALSE
    )

Duplicate DOIs in the Bibliography
Type	Identifier
doi	10.1111/emr.12464
doi	10.1111/mam.12277

DOIs in Open Alex

To validate the existence and validity of the DOIs, we check if the DOIs are in the OpenAlex database.

Show the code

dois_not_in_oa <- unique(dois)[!(unique(dois) %in% dois_oa)]

dois_valid <- dois_not_in_oa[IPBES.R::doi_valid(dois_not_in_oa)]

Of the 3861 in the library which have a DOI, 471 (12.1989122%) are in not in OpenAlex.

Show the code

data.frame(
    Type = "doi",
    Identifier = sprintf('<a href="https://doi.org/%s" target="_blank">%s</a>', dois_not_in_oa, dois_not_in_oa)
) |>
    IPBES.R::table_dt(caption = "The Caption")

Of these 3 are not valid. These are:

Show the code

data.frame(
    Type = "doi",
    Identifier = sprintf('<a href="https://doi.org/%s" target="_blank">%s</a>', dois_not_in_oa[!(dois_not_in_oa %in% dois_valid)], dois_not_in_oa[!(dois_not_in_oa %in% dois_valid)])
) |>
    knitr::kable(
        caption = "Non Valid DOIs in the Bibliography",
        escape = FALSE
    )

Non Valid DOIs in the Bibliography
Type	Identifier
doi	annurev.es.23.110192.000431
doi	10.1371/ journal.pone.0111913
doi	https://www.sciencedirect.com/science/article/abs/pii/S0169534709002018

TODO Finally we check, if these dois exist but are not ingested into OpanAlex. This is done using the doi.org resolver This is disabled at the moment.

Show the code

dois_exist <- IPBES.R::doi_exists(
    dois_valid,
    cache_file = file.path(".", "cache", "doi_exist.rds")
)

Show the code

to_check <- dois[!(dois %in% dois_oa)]

dois_valid <- IPBES.R::doi_valid(dois)
dois_openalex <- dois %in% dois_oa
names(dois_openalex) <- dois

dois_exist <- IPBES.R::doi_exists(to_check, cache_file = file.path(".", "cache", "doi_exist.rds"))
dois_not_retracted <- IPBES.R::doi_not_retracted(dois, cache_file = file.path(".", "cache", "doi_not_retracted.rds"))

sprintf(
    fmt = paste(
        "Number of references: \t\t %d",
        "Number of DOIs: \t\t %d",
        "Number of Duplicate DOIs: \t %d",
        "Number of DOIs in OpenAlex: \t %d ( %f %)",
        "Number of Existing DOIs: \t %d",
        "Number of Retracted DOIs: \t %d",
        "Percentage of Duplicate DOIs: \t %f",
        sep = "\n"
    ),
    nrow(bib),
    sum(!is.na(dois)),
    length(dois) - length(unique((dois))),
    sum(dois_openalex), 100 * sum(dois_openalex) / nrow(bib),
    sum(dois_exist),
    sum(!dois_not_retracted),
    ((dois_valid |> unique() |> length()) / length(dois_valid)) |> round(digits = 3) * 100
) |> cat()

Show the code

oldopts <- options(knitr.kable.NA = "")
data.frame(
    Measure = c(
        "# References",
        "**DOI**",
        "# DOIs",
        "# Duplicate DOIs",
        "# Existing DOIs",
        "# Retracted DOIs",
        "% Duplicate DOIs",
        "**ISBN**",
        "# ISBNs",
        "# Duplicate ISBNs",
        "**ISSN**",
        "# ISSNs",
        "# Duplicate ISSNs"
    ),
    Value = c(
        nrow(bib),
        NA,
        sum(!is.na(dois)),
        length(dois) - length(unique((dois))),
        sum(dois_exist),
        sum(!dois_not_retracted),
        ((dois_valid |> unique() |> length()) / length(dois_valid)) |> round(digits = 3) * 100,
        NA,
        sum(!is.na(isbns)),
        length(isbns) - length(unique((isbns))),
        NA,
        sum(!is.na(issns)),
        length(issns) - length(unique((issns)))
    )
) |>
    knitr::kable(
        caption = "Cleanliness of the Bibliography",
    )
options(oldopts)

Contentual and Bibliographic analysis

Publication types

Show the code

bib |>
    dplyr::group_by(
        Item.Type
    ) |>
    dplyr::summarize(
        count = n()
    ) |>
    dplyr::arrange(
        desc(count)
    ) |>
    knitr::kable()

Item.Type	count
journalArticle	4157
bookSection	305
book	244
report	236
webpage	97
conferencePaper	83
document	67
thesis	15
dataset	12
blogPost	9
preprint	2
magazineArticle	1
newspaperArticle	1
presentation	1
videoRecording	1

Year of Publication

Show the code

data_bib <- bib |>
    dplyr::group_by(
        Publication.Year,
        Item.Type
    ) |>
    dplyr::summarize(
        count = n(),
        p = count / sum(count),
    ) |>
    dplyr::group_by(
        Item.Type
    ) |>
    dplyr::arrange(
        Publication.Year
    ) |>
    mutate(
        count_cumsum = cumsum(count),
        p_cumsum = cumsum(p)
    ) |>
    dplyr::rename(
        publication_year = Publication.Year,
        type = Item.Type
    ) |>
    dplyr::mutate(
        type = dplyr::case_match(
            type,
            "journalArticle" ~ "article",
            .default = type
        )
    )

`summarise()` has grouped output by 'Publication.Year'. You can override using
the `.groups` argument.

Show the code

data_works <- bib_works |>
    dplyr::group_by(
        publication_year,
        type
    ) |>
    dplyr::summarize(
        count = n(),
        p = count / sum(count),
    ) |>
    dplyr::group_by(
        type
    ) |>
    dplyr::arrange(
        publication_year
    ) |>
    mutate(
        count_cumsum = cumsum(count),
        p_cumsum = cumsum(p)
    ) |>
    dplyr::rename(
        publication_year = publication_year,
        type = type,
        count_oa = count,
        p_oa = p,
        count_oa_cumsum = count_cumsum,
        p_oa_cumsum = p_cumsum
    )

`summarise()` has grouped output by 'publication_year'. You can override using
the `.groups` argument.

Show the code

data <- dplyr::full_join(
    x = data_bib,
    y = data_works,
    by = c("publication_year", "type")
)

rm(data_bib, data_works)

data |>
    dplyr::filter(publication_year >= 1950) |>
    ggplot() +
    scale_fill_viridis_d(option = "plasma") +
    geom_line(aes(x = publication_year, y = count_cumsum / 10, colour = type), linetype = "solid") + # Zotero
    geom_line(aes(x = publication_year, y = count_oa_cumsum / 10, colour = type), linetype = "dashed") + # OpenAlex
    scale_x_continuous(
        breaks = seq(1500, 2020, 10)
    ) +
    scale_y_continuous(
        "Proportion of publications",
        sec.axis = sec_axis(~ . * 10, name = "Cumulative number of references") # divide by 10 to scale back the secondary axis
    ) +
    labs(
        title = "Publications over time",
        x = "Year",
        y = "Number of publications"
    ) +
    theme_minimal() +
    theme(axis.text.y.right = element_text(color = "red")) +
    theme(legend.position = "bottom") +
    guides(
        fill = guide_legend(
            title = "Legend"
        )
    )

Warning: Removed 11 rows containing missing values or values outside the scale range
(`geom_line()`).

Warning: Removed 198 rows containing missing values or values outside the scale range
(`geom_line()`).

Access Status of References

This is checked by using the OpenAlex retrieved works. Therefore it is li=mited to the works that are on OpenAlex. At the moment, only references with a DOI were retrieved from OpenAlex.

Show the code

bib_works |>
    ggplot(
        aes(
            x = publication_year,
            fill = oa_status
        )
    ) +
    geom_bar(
        position = "fill"
    ) +
    scale_fill_manual(values = c("#CD7F32", "red", "gold", "green", "pink")) +
    ggtitle("Publication Year") +
    theme(
        plot.title = element_text(size = 15)
    ) +
    theme(legend.position = "bottom")

50 Most often cited Journals

Show the code

data <- bib |>
    dplyr::group_by(
        Publication.Title
    ) |>
    dplyr::summarise(
        count = n()
    ) |>
    dplyr::rename(
        Journal = Publication.Title
    ) |>
    arrange(
        desc(count)
    )

data |>
    dplyr::slice_max(
        count,
        n = 50
    ) |>
    ggplot(
        aes(
            x = reorder(Journal, -count),
            y = count
        )
    ) +
    geom_bar(
        stat = "identity",
        fill = "steelblue"
    ) +
    coord_flip() +
    labs(
        x = "Journal",
        y = "Count"
    ) +
    ggtitle(
        "Most often cited Journals"
    ) +
    theme(
        plot.title = element_text(size = 15)
    )

This table contains all Journals as specified in the Zotero database.

Show the code

IPBES.R::table_dt(data, "cited_journals")

TODO Coutries of Institutes of all authors

sapply(bib_works$author, function(x){x[“institution_country_code”]}) |> unlist() |> table() |> sort()

Show the code

bib_works |>
    dplyr::group_by(
        institution_country_code
    )


top_countries <- sapply(bib_works$author, function(x) {
    x["institution_country_code"]
}) |>
    unlist() |>
    table() |>
    sort(decreasing = TRUE) |>
    head(50)

data.frame(
    Country = names(top_countries),
    Count = as.numeric(top_countries)
) |>
    ggplot(
        aes(
            x = reorder(Country, -Count),
            y = Count
        )
    ) +
    geom_bar(
        stat = "identity",
        fill = "steelblue"
    ) +
    coord_flip() +
    labs(
        x = "Country",
        y = "Count"
    ) +
    ggtitle(
        "Countries of Institutes of all Authors"
    ) +
    theme(
        plot.title = element_text(size = 15)
    )

VOSViewer Example

TODO ILK References in Bibliography

Show the code

#|

ilk_id <- list(
    doi = ilk$DOI[dois_valid] |>
        unique(),
    isbn = ilk$ISBN[ilk$ISBN != ""] |>
        unique(),
    issn = ilk$ISSN[ilk$ISSN != ""] |>
        unique()
)

# ilk_snowball_id <- list(
#     doi = ilk_snowball$nodes$doi[ilk_snowball$nodes$doi != ""] |>
#         unique()
# )

va <- read.csv(file.path("input", "./Values Assessment.csv"))

va_id <- list(
    doi = va$DOI[va$DOI != ""] |>
        unique(),
    isbn = va$ISBN[va$ISBN != ""] |>
        unique(),
    issn = va$ISSN[va$ISSN != ""] |>
        unique()
)

In the ILK database, there are nrow(ilk) references, while in the Value Assessment bibliography has nrow(va) references.

We now determine the overlap of these two by comparing the DOIs.

Show the code

#|

# both_id <- list(
#     doi = va_id$doi[(va_id$doi %in% ilk_id$doi)],
#     isbn = va_id$isbn[(va_id$isbn %in% ilk_id$isbn)],
#     issn = va_id$issn[(va_id$issn %in% ilk_id$issn)] # ,
#     #     doi_snowball = va_id$doi[(va_id$doi %in% gsub("https://doi.org/", "", ilk_snowball_id$doi))]
# )

# count <- sapply(both_id, length)
# va_ref <- sapply(va_id, length)
# va_ref <- c(va_ref, doi_snowball = va_ref[[1]])

# data.frame(
#     ref_ILK = length(),
#     va_ref = va_ref,
#     p = round(count / va_ref, digits = 3)
# ) |>
#     knitr::kable(
#         caption = "Overlap of the ILK and the Values Assessment Bibliography",
#         col.names = c("ILK ref in VA", "n VA ref", "p")
#     )


data.frame(
    No.ILK.Ref = nrow(ilk),
    No.Valid.DOIs = length(dois_valid),
    No.VA.Ref = nrow(va),
    No.DOI.both = sum(dois_valid %in% va$DOI),
    No.ISBN.both = sum((ilk$ISBN[ilk$ISBN != ""]) %in% va$ISBN),
    No.ISSN.both = sum((ilk$ISSN[ilk$ISSN != ""]) %in% va$ISBN)
) |> knitr::kable(
    caption = "Number of references in ILK Database"
)

TODO TO BE REPLACED WITH TOPICS Assessment of OpenAlex Concepts

Show the code

#|


concepts <- do.call(rbind, ilk_works[["concepts"]])

concepts_summary <- concepts |>
    #     filter(level == 1) |>
    mutate(score = round(score * 100)) |>
    group_by(display_name, level) |>
    summarize(score = mean(score), count = n()) |>
    arrange(desc(count))

Show the code

#|

ggraph(graph = as_tbl_graph(ilk_snowball), layout = "stress") +
    geom_edge_link(aes(alpha = after_stat(index)), show.legend = FALSE) +
    geom_node_point(aes(fill = oa_input, size = cited_by_count), shape = 21, color = "white") +
    geom_node_label(aes(filter = oa_input, label = id), nudge_y = 0.2, size = 3) +
    scale_edge_width(range = c(0.1, 1.5), guide = "none") +
    scale_size(range = c(3, 10), guide = "none") +
    scale_fill_manual(values = c("#a3ad62", "#d46780"), na.value = "grey", name = "") +
    theme_graph() +
    theme(
        plot.background = element_rect(fill = "transparent", colour = NA),
        panel.background = element_rect(fill = "transparent", colour = NA),
        legend.position = "bottom"
    ) +
    guides(fill = "none")

Introduction

Bibliography Setup

Data Quality of the Bibliography

Cleanliness of bibliography

Entries with DOIs, ISBNs or ISSNs

DOIs in Open Alex

Contentual and Bibliographic analysis

Publication types

Year of Publication

Access Status of References

50 Most often cited Journals

TODO Coutries of Institutes of all authors

VOSViewer Example

TODO ILK References in Bibliography

TODO TO BE REPLACED WITH TOPICS Assessment of OpenAlex Concepts